Framework
Discussion
Temporal Modified Speech Perception in Dyslexia
Contacts : [email protected], [email protected]
Group effect. Dyslexic adults exhibit more difficulties to identify the temporal modified speech for both phonetic features (voicing and place of articulation).In addition, our results revealed a large inter-individual variability in each group. This variability was also observed in a previous study using reversed speech stimuli (Meunier et al., 2002). Degraded speech identification may differ according subject’s cognitive abilities.
Phonetic features. Stronger deficit of integration for the VOTthan for the second formant transition. In Experiment 1, the attack and the intervocalic consonants were affected whereas in Experiment 2, only the attack consonant was sensible to a group effect. In accordance with the strongly effect of voicing observed in Ziegler et al.’s study (2005).
Auditory theory. Time-compressed non-word auditory identification deficit for dyslexic adults compared to control adults. Dyslexics may show an auditory deficit in temporal processing and particularly in time-compressed speech. In repetition task for dyslexics: they only have to decode stimuli at the first acoustic stage (acoustic features extraction). In opposition with Rosen’s opinion (2003): the rapid auditory processing would be impaired with dyslexia.
To conclude.Specific identification deficit on consonant even at normal rate for dyslexics and it is worst and worst with the speech degradation. The temporal organisation seems to play an important role for dyslexics.
In addition of this temporal processing deficit, we assume that cognitive restoration capacities of dyslexics are also deficient. The descendant auditory pathway might be a cognitive feedback to modulate the temporal resolution. Further hearing tests in order to evaluate the central auditory system of dyslexic adults would inform us on their central
auditory abilities. Literature
Foulke, E. (1971). The perception of time-compressed speech. In D. L. Horton & J. J. Jenkins (Eds.), The perception of language. Columbus, Ohio: Charles E. Merrill publishing company.
Jacquier-Roux, M., Valdois, S., & Zorman, M. (2002) Outils de dépistage des dyslexies. Académie de Grenoble: Laboratoire Cogni-Sciences et Laboratoire de Psychologie et Neuro-cognition.
Janse, E. (2003). Production and perception of fast speech. Doctoral dissertation, University of Utrecht.
Lefavrais, P. (1965). Test de l’Alouette. Paris: E.C.P.A.
Merzenich, M. M., Jenkins, W. M., Johnston, P., Schreiner, C., Miller, S. L., & Tallal, P. (1996). Temporal processing deficits of language-learning impaired children ameliorated by training. Science, 271, 77-81.
Meunier, F., Cenier, T., Barkat, M., & Magrin-Chagnolleau, I. (2002). Mesure d’intelligibilité de segments de parole à l’envers en français. In proc. of XXIVèmes Journées d’Etude sur la Parole, pp. 117-120.
Rosen, S. (2003). Auditory processing in dyslexia and specific language impairment. Is there a deficit? What is its nature? Does it explain anything? Journal of Phonetics, 31, 509–527.
Tallal, P. (1980). Auditory temporal perception, phonics, and reading disabilities in children, Brain and language, 9, 182-198.
Tallal, P., Galaburda, A. M., Llinás, R. R. & von Euler, C. (Eds.) (1993). Temporal information processing in the nervous system. New York: New York Academy of Sciences.
Tallal, P., Miller, S. L., Bedi, G., Byma, G., Wang, X., Nagarajan, S. S., Schreiner, C., Jenkins, W. M., & Merzenich, M. M. (1996). Language comprehension in language-learning impaired children improved with acoustically modified speech. Science, 271, 81-84.
Ziegler, J. C., Pech-Georgel, C., George, FX., Alario, F. & Lorenzi, C. (2005). Deficits in speech perception predict language learning impairment. Proceedings of the National Academy of Sciences, pp. 14110-14115.
Participants
- Maternal language : French.
- Normal hearing.
Method
Manual segmentation VOT and FT2 Time-Compression for each acoustic cue
according to 4 conditions :• 100% = original duration
• 50% = 50% of original duration left
• 25% = 25% of original duration left
• 0% = totally deleted
Stimuli
- Male French speaker
- 64 disyllabic non-words C1V1C2V2
- 16 fillers V1C1V2
Subjects’ Task
Non-words binaural auditory identification.
• Control group: transcription task
• Dyslexic group: repetition task
b a b a
d i d i
p p
t t
Figure 1. Schematic illustration of non-words structure.
•Auditory hypothesis(Tallal, 1980).
• General auditory deficit in temporal processing (Tallal et al., 1993).
• Training of language impaired children with temporally stretched speech (Merzenich et al., 1996; Tallal et al., 1996).
• Accelerated Speech Intelligibility and Perception?
• The temporal organisation of spoken language may become more critical to comprehension, the more the speech rate is increased (Foulke, 1971).
Artificially time-compressed speech was easier to process than naturally produced fast speech (Janse, 2003). Hypothesis: an acceleration of the speech signal rate would change the speech perception of dyslexic adults. Dyslexics would have more difficulties to process artificially time-compressed speech than controls.
Rapid transition and brief sound of the speech signal: acoustic cues are essential in speech perception, the Voice Onset Time (VOT) and the transition of the formant 2 (FT2). Hypothesis: dyslexics would show more or less specific impairment for each acoustic cue compared to controls.
Caroline Jacquier & Fanny Meunier
Laboratoire Dynamique du Langage, UMR 5596, CNRS & Université Lumière Lyon 2
22 22
22 22
female
10 10
10 10
male Gender
23.3 22.6
23.3 22.6
Mean Age (yrs)
Dyslexic adults (N = 32) Control adults
(N = 32) Dyslexic adults
(N = 32) Control adults
(N = 32)
Experiment 2 Experiment 1
Method
Experiment 1: VOT time-compression
Vowels are better identified than consonants.
3-way ANOVA
Group effect(p < .001).
Position effect(p < .001).
Time-compression effect(p < .001).
Group x Position Interaction (p < .05).
Position x Time-Compression Interaction (p < .01).
Inter-individual variability.
Results
Experiment 2: FT2 time-compression
VOT Experiment
50%
60%
70%
80%
90%
100%
110%
C1 V1 C2 V2 Non
Word
Identification rate (%)
Dyslexic Control
FT2 Experiment
50%
60%
70%
80%
90%
100%
110%
C1 V1 C2 V2 Non
Word
Identification rate (%)
Dyslexic Control
*** ***
***
*** *** Vowels are better identified than
consonants.
3-way ANOVA
Group effect(p < .01).
Position effect(p < .001).
Time-compression effect(p < .001).
Group x Position Interaction (p < .05).
Position x Time-Compression Interaction (p < .001).
Results
/b/ /i/ /p/ /a/
- Dyslexic adults vs. Control adults.
- Dyslexia diagnosed+ French dyslexia detection test (ODEDYS created by Jacquier-Roux, Valdois, & Zorman, 2002) and Alouette test (Lefavrais, 1965).
Table 1. Participant groups in experiment 1 and 2.
Figure 2. Spectrogram of non-word [bipa], time-compression 50%.
Figure 3. Histograms of the identification rate (%) of each
phoneme (C and V) and the non word. Figure 4. Histograms of the identification rate (%) of each phoneme (C and V) and the non word.
Confusion Matrix
• Attack consonant: confounded with the unvoiced corresponding consonant (/b/->/p/ or /d/->/t/) even at 100%
(control condition) for dyslexics.
• Intervocalic consonant: Voiced consonants (/b/ and /d/) confounded with the approximant liquid consonant /l/.
Confusion Matrix
• A main type of confusion: a place of articulation error for voiced consonants (/b/->/d/) in both consonant positions for both groups but appears as from the control condition (100%) for dyslexics.
• It is noteworthy that dyslexic adults present significant auditory non- words identification deficit in the control condition in which the speech signal is intact.
Table 2. Identification rates (%) for VOT and FT2 compression experiments for Control vs. Dyslexic groups.
FT2 compression Dyslexic adults SD
95.3 97.3 67.8
98.4 94.7
98.2 99
80.5 92 70.1
92.4 92.8
93.9 95.5
0%
25%
50%
0%
25%
50%
100%
100%
C2 C1
95.7 99.6 99.4 99.6 85.2 96.5 98.2 97.5 Control adults
0.17 0.09 0.03 0.05 0.13 0.10 0.09 0.09 SD
62.9 87.3 97.7 97.5 59.2 85.2 87.9 91.2 Dyslexic adults
0.04 0.14 0.04
0.04 0.06 0.02
0.03 0.02 0.02
0.03 0.02 0.02
99.6
0.11 0.10 0.08
0.10 0.06 0.04
97.3 Control adults
VOT compression
0.04 0.04
0.09 0.04
0.09 0.04
SD SD
